Skip to main content
View rawEdit

Dashboards

Dashboards are the deep-dive layer of postgres_ai monitoring. Use them when Checkup tells you what is wrong and you need to understand why or how bad it is right now.

When to use dashboards​

  • Checkup reports high query time — inspect query patterns and waits
  • Checkup reports table bloat — verify growth, vacuum history, dead tuples
  • The app is slow right now — find active sessions, waits, and blockers
  • You need real-time visibility during an incident
  • Weekly health review to catch trends before they become problems

Dashboard list​

#DashboardUse for
01Node overviewFirst stop: sessions, throughput, waits, resource usage
02Query performance (top-N)Identify the heaviest queries by time, calls, or I/O
03Single query analysisDeep dive into one query's performance over time
04Wait event analysis (ASH)Breakdown of what sessions are waiting on
05Backups and DRBackup health, WAL generation, recovery readiness
06Replication and HAReplication lag, slot health, HA posture
07Autovacuum and bloatVacuum activity, dead tuples, bloat trends
08Aggregated table analysisCompare all tables by size, growth, and maintenance
09Single table analysisDeep dive into one table's stats and vacuum history
10Aggregated index analysisFind unused, redundant, or bloated indexes
11Single index analysisVerify usage and health of a specific index
12SLRU cache statsTransaction and commit log cache performance
13Lock contentionIdentify blocking queries and lock wait patterns

Investigation workflows​

"My app is slow right now" (10-15 minutes)​

  1. Node overview — look at active sessions and wait distribution
  2. Wait events — identify the dominant wait type (IO, Lock, CPU)
  3. Lock contention — if locks dominate, find blocking queries
  4. Query performance — identify top resource consumers
  5. Single query — drill into the worst query's timeline

"Checkup found unused indexes" (5-10 minutes)​

  1. Aggregated index analysis — sort by zero scans / low usage
  2. Single index analysis — confirm no scans, check size and table context
  3. Query performance — ensure no critical queries depend on the index
  4. Drop with confidence

"Checkup found table bloat" (5-10 minutes)​

  1. Aggregated table analysis — find the bloated table and dead tuple ratio
  2. Single table analysis — review growth trend and vacuum history
  3. Autovacuum and bloat — check worker activity and lag
  4. Fix with VACUUM, tuning, or pg_repack

"Weekly health review" (10-15 minutes)​

  1. Node overview — any anomalies in the last week?
  2. Aggregated table analysis — unexpected growth or dead tuples?
  3. Autovacuum and bloat — keeping up across the biggest tables?
  4. Aggregated index analysis — new unused or bloated indexes?

Accessing dashboards​

Local installation​

npx postgresai mon local-install --demo
# Open http://localhost:3000

Install from PostgresAI Console for a fully managed deployment. You get dashboards, daily health check reports with 23 checks, issue tracking, and AI-assisted resolution — without managing Docker infrastructure yourself.

Live demo​

https://demo.postgres.ai (login: demo / password: demo)